Youth Violence and Juvenile Justice Race, Poverty, and The

Article

Race, Poverty, andExclusionary SchoolSecurity: An EmpiricalAnalysis of U.S. Elementary,Middle, and High Schools

Aaron Kupchik1 and Geoff Ward2

AbstractAs violence and crime within and around U.S. schools has drawn increased attention to schoolsecurity, police, surveillance cameras, and other measures have grown commonplace at publicschools. Social scientists commonly voice concern that exclusionary security measures are mostcommon in schools attended by poor and non-White students, yet there is little empirical basis forassessing the extent of differential exposure, as we lack research on how exclusionary measures aredistributed relative to school and student characteristics. To address this gap in the research, we usenationally representative school-level data from the School Survey on Crime and Safety to considerthe security measures employed in elementary, middle, and high schools. Results indicate that whilesecurity measures are ubiquitous in U.S. high schools, those considered more exclusionary are con-centrated in elementary, middle, and high schools attended by non-White and/or poorer students.

Keywordsschool security, racial inequality, class inequality, governing through crime

The tragic shooting deaths of 20 children and 6 adult staff members at Sandy Hook Elementary

School in 2012 have drawn renewed attention to the issue of school security. Although debates rage

about gun control, arming school personnel, and other proposals for reducing violence in schools and

beyond, it is important for social science research to consider the scope and significance of extensive

school security interventions already in place. Indeed, schools across the United States have incor-

porated a host of security mechanisms over the past decade. According to the National Center for

Education Statistics, in the 2009–2010 school year, 60% of public high schools conducted random

searches using drug-sniffing police dogs, 84% used surveillance cameras, and 12% used metal

1 University of Delaware, Newark, DE, USA2 University of California, Irvine, CA, USA

Corresponding Author:

Aaron Kupchik, University of Delaware, 329 Smith Hall, Newark, DE 19716, USA.

Email: [email protected]

Youth Violence and Juvenile Justice2014, Vol. 12(4) 332-354ª The Author(s) 2013Reprints and permission:sagepub.com/journalsPermissions.navDOI: 10.1177/1541204013503890yvj.sagepub.com

at UNIV OF DELAWARE LIB on September 22, 2014yvj.sagepub.comDownloaded from

http://www.sagepub.com/journalsPermissions.nav

http://yvj.sagepub.com

http://yvj.sagepub.com/

detectors to screen students. In a 2009 nationally representative survey of students aged 12–18, 68%reported that a security guard or police officer was assigned to their school (Robers, Zhang, Truman,

& Snyder, 2012).

Although the proliferation of school security measures is clear, we know little about variation in

specific security strategies across school contexts, including how exposure to more protective (i.e.,

inclusive) versus punitive (exclusive) measures relates to school, student, and community character-

istics. Indeed, while public and political discourse on school security tends to emphasize student and

staff safety, many social scientists have voiced concern that more exclusionary measures may do less

to protect students than to compromise educational environments and outcomes (e.g., Fabelo et al.,

2011; Hirschfield, 2008, 2010; Kupchik, 2010; Nolan, 2011). Insofar as more protective inclusion-

ary measures are concentrated in schools serving White and affluent student populations, and puni-

tive exclusionary measures distinguish schools attended by the poor and non-Whites, as these critics

contend, the race and class stratification of school security measures may work to perpetuate

inequality (Hirschfield, 2010; Irwin, Davidson, & Hall-Sanchez, 2013; Wacquant, 2001). The exist-

ing research makes it abundantly clear that racial/ethnic minority youth are significantly more likely

to be subjected to school punishment than White students (e.g., Eitle & Eitle, 2004; Fabelo et al.,

2011; Skiba, Michael, Nardo, & Peterson, 2000; Skiba et al., 2006), though these effects are seen

at the individual level, largely due to biased perceptions of school staff (e.g. Bowditch, 1993;

Downey & Pribesh, 2004; Ferguson, 2000; Morris, 2005), and leave us with little understanding

of variations in policies across schools.

As we discuss further below, researchers point out that school security may reflect an ethos of

exclusion, an ethos of inclusion, or some combination of both (Hirschfield, 2010). Exclusionary

measures uniquely aim to disengage offending students from school. Exclusionary measures may

also repel other students, facilitating broader disintegration of larger student bodies from such school

environments, and socioeconomic mainstreams (e.g. labor markets) connected to education. In 2011,

the Los Angeles Unified School District and its police department acknowledged this exclusion

when vowing to roll back punitive measures. ‘‘The old rules were part of a get-tough philosophy

that included truancy sweeps, $250 tickets and mandatory court appearances that could result in jail

time for parents,’’ the Los Angeles Times reports. The district faced mounting concern that the mea-

sures ‘‘diminish time in school and increase the dropout rate’’ (Blume, 2011).

Although research finds that firm rule enforcement is important both for student safety and for

academic outcomes (Arum & Velez, 2012), a growing body of literature illustrates how an overre-

liance on exclusionary security may contribute to school disengagement and social disintegration in

several interrelated ways. Exclusionary security measures deteriorate the school social climate

(Ayers, Dohrn, & Ayers, 2001; Brady, Balmer, & Phenix, 2007; Lewis, Romi, Katz, & Qui,

2008; Lyons & Drew, 2006; Webber, 2003), which in turn can lead to reduced student academic

performance and increased misbehavior (Gottfredson, 2001; Gottfredson, Gottfredson, Payne, &

Gottfredson, 2005; Welsh, 2003). These adverse impacts on learning environments and experiences

may increase exposure to further school discipline and the likelihood of dislocation and exclusion

from school through suspension, dropout, and expulsion (Ayers et al., 2001; Fabelo et al., 2011;

Hirschfield, 2008; Kim, Losen, & Hewitt, 2010; Skiba et al., 2000, 2006). At its extreme, this may

yield what some call a ‘‘school to prison pipeline,’’ where students are diverted from schools to

courts and carceral institutions via the criminalization of school misconduct (Kim et al., 2010; Wald

& Losen, 2003). If distinct to schools attended by non-White and poorer students, the denial of

opportunities for educational and social growth that result from exclusionary security would contrib-

ute to the reproduction of inequality.

These concerns raise a host of empirical and policy questions that cannot be addressed in their

entirety here. Instead, we take on a narrower but fundamental question raised by the suggestion that

exclusionary school security measures are contributing to the reproduction of inequality: the

Kupchik and Ward 333



question of whether in fact more exclusionary security measures—as others have defined them—are

concentrated in schools serving non-White and poorer students. Although this assertion has taken on

an almost common sense quality at times, the scarcity of empirical research on school-level differ-

ences across the United States leaves the answer unclear. Further, among existing studies of school-

level differences, there is disagreement over whether the rise of punitiveness in school discipline is a

general phenomenon, reflective of a new ‘‘law and order governmentality,’’ or concentrated in

schools serving otherwise marginalized populations (see Irwin et al., 2013).

Drawing on summary data or small samples of schools, some researchers emphasize a general

trend in the expansion of school security that is facilitated by federal policy and funding. In response

to federal mandates in the 1990s, for example, some form of ‘‘zero tolerance’’ is seemingly universal

in U.S. public high schools (Simon, 2007; also see Kupchik, 2010). Further, federal subsidies for

state and local expenditures on school security have contributed to apparently widespread adoption

of school police officers (usually referred to as school resource officers, or SROs), surveillance cam-

eras, and other measures (see Hirschfield, 2010). In a qualitative comparison of four high schools,

two of which consist of mostly middle-class White students and two primarily comprising lower

income youth of color, Kupchik (2009, 2010) reports that all of the schools adopted similar harsh,

exclusive strategies mirroring criminal justice practices. Although noting some differences in disci-

plinary practices, he finds a governmentality of security and punishment encompassing schools

across social strata. However, researchers also suggest that the expansion of school security has been

more selective, targeting certain schools and students with more punitive exclusionary measures,

potentially reproducing inequality in educational and associated outcomes (Hirschfield, 2008; Irwin

et al., 2013; Payne & Welch, 2010; Wacquant, 2001; Welch & Payne, 2010, 2012).

Very few studies offer empirical tests of whether student social status (measured at the school

level) relates to school punishment. In one set of studies, Payne and Welch (2010) and Welch and

Payne (2010, 2012) analyze a nationally representative sample of public middle schools and high

schools to determine whether the race and socioeconomic status of students is related to the harsh-

ness of school punishment. They find that harsh punishment (measured as a scale) is indeed more

likely, and mild or restorative punishment less likely, in schools with larger proportions of Black

and (to a lesser extent) poor students (Welch & Payne, 2010); that schools using one type of harsh

punishment tend to use others as well (Payne & Welch, 2010); and that the proportion of Black stu-

dents relates to the use of specific punishment practices such as expulsion, suspension, and in-school

suspension (Welch & Payne, 2012). These studies greatly improve our understanding of punitive-

ness in schools, but they focus solely on punishments rather than on security measures intended

to prevent student misconduct and keep youth safe. Security practices are important to study as well.

Interests in securing schools are longstanding and widespread and have gained renewed focus in the

post-Newtown debate on school safety. Beyond research on punishment (e.g., expulsion) in response

to specific acts of misconduct, we must also examine the built environments of school security (e.g.,

the presence or absence of police, lining up to pass through metal detectors, etc.), including their

potentially inclusive and exclusive dynamics.

In a recent analysis, Irwin, Davidson, and Hall-Sanchez (2013) use nationally representative data

(the SSOCS survey, which we use as well) to assess whether race and poverty relate to both punish-

ment and security across schools. Here punishment is operationalized in two ways, using measures

of various punishments given in response to misbehavior and the number of crimes schools report to

the police; security is measured by the number of weekly hours worked at the school by either secu-

rity guard or law enforcement personnel and by the number of surveillance mechanisms employed

by each school. They find that the presence of racial/ethnic minority youth is positively related to the

use of school punishments, crimes reported to the police, and security or police presence but not to

surveillance practices. Schools with more poor youth use fewer punishments, rely less on security

guards and police, and report fewer crimes to the police but have more surveillance practices.

334 Youth Violence and Juvenile Justice 12(4)



Irwin et al.’s (2013) analyses offer the first test of how student body racial composition and pov-

erty rates shape security practices. Yet one result of the breadth of their work—they consider both

punishment and security, across all school levels, and in ways that aggregate distinct security prac-

tices—is that we still do not know how race and poverty relate to the use of specific security stra-

tegies, including more exclusive or inclusive approaches. Our analyses follow this study closely and

build on it in a particular direction. Rather than considering both punishment and security, we focus

particularly on school security in order to disentangle the use (and predictors of use) of distinctly

exclusionary security measures, and we consider how these relationships vary across school levels.

This is an important component to our research, since prior work tends to be based either on high

schools only (e.g, Kupchik, 2010; Lyons & Drew, 2006) or on multiple levels of schools aggregated

together (e.g., Payne & Welch, 2010; Welch & Payne, 2010, 2012).

This article thus builds on prior research with an empirical analysis of how race and poverty relate

to exposure to exclusive school security measures across U.S. elementary, middle, and high schools.

We employ Hirschfield’s (2008, 2010) distinction between an ‘‘inclusionary versus exclusionary

ethos’’ of school security to assess how race and socioeconomic composition of students relates

to the presence of exclusionary security measures, net of school and community crime, and other

contextual factors likely to influence the school security environment. We draw on two theoretical

frameworks, offering opposing predictions of differential exposure to exclusive school security: a

social reproduction perspective, anticipating disparate or selective exclusion, and a ‘‘governing

everyone through crime’’ perspective (see also Irwin et al., 2013).

Background: Securing the American School

Public and political discourse present the investment in school security as an essential response to

high levels of crime within and around schools, using previous incidents and the threat of future

crime and violence to justify new security and disciplinary measures (see Brooks, Schiraldi, & Zie-

denberg, 2000; Burns & Crawford, 1999; Cornell, 2006; Lawrence & Mueller, 2003; Simon, 2007).

Federal, state, and local government efforts have responded to concerns about school crime with a

range of interventions. Since 2000, Congress has committed around US$15 million per year to the

national ‘‘Secure Our Schools’’ (SOS) Act, an amendment to the Omnibus Crime Control and Safe

Streets Act of 1968. SOS is a voluntary, matching grant program where municipalities apply for fed-

eral funding for school safety grants administered under the Community-Oriented Policing Services

(COPS) program. SOS grants can be used for metal detectors, locks, lighting, deterrent measures,

security assessments, and security training, with the federal government paying half of the cost of

security measures and state or local government providing remaining portions. Similar funding has

been provided through the Department of Justice, the Department of Education, the Department of

Homeland Security, and state governments (Casella, 2006; Lohman & Shephard, 2006; U.S. Con-

gress 2000; U.S. Department of Justice 2011).

Often framed as school and student ‘‘accountability’’ programs, these interventions encourage

aggressive security monitoring, classification, and intervention and contribute to broader efforts

to reorganize school systems. The 1994 Safe and Drug-free Schools Act, for example, encouraged

criminal classification of school misconduct and increased school surveillance by tying block grants

to demonstrate school crime problems and encouraging that funds be used to enhance surveillance

(Simon, 2007). The No Child Left Behind Act of 2001 includes a provision requiring states to work

with school districts to identify ‘‘persistently dangerous schools’’ and mandates criminal referrals for

students in possession of a weapon (Hirschfield, 2010; Monahan & Torres, 2010). Local officials

have moved to exert greater control over schools identified as underperforming and unsafe. For

example, in New York City, authority over school safety agents was shifted from educational




officials to the New York Police Department, increasing the aggressiveness of policing and likeli-

hood of criminal sanctions in the event of school misconduct (Hirschfield, 2010).

Despite the political framing of school security as a response to rampant violence and crime,

school crime has been decreasing over the past 20 years (Robers et al., 2012). Sociologists and crim-

inologists have instead theorized the rise in school security to be a consequence of phenomena such

as racial threat (see Payne & Welch, 2010; Welch & Payne, 2010, 2012), neoliberal strategies for

governing students (see Hirschfield, 2008; Kupchik & Monahan, 2006), reduced moral authority and

fear of litigation among schools (Arum, 2003), and broader contemporary strategies for managing

risk and insecurity, such as ‘‘governing through crime’’ (Simon, 2007). Research into the application

of these strategies within schools finds that they are often excessive; though firm and fair discipline

and security are necessary to protect students and maintain order, contemporary practices in some

schools are far more rigid and punitive than is required for safe and orderly environments (see Nolan,

2011). Moreover, because they exclude larger numbers of students than necessary from opportuni-

ties for educational and social growth, while punishing youth rather than addressing the underlying

reasons for student misbehavior, contemporary school security may have adverse consequences for

educational environments and outcomes (e.g., Casella, 2001; Kupchik, 2010; Nolan, 2011). If these

detrimental effects are more pronounced in schools attended by poorer and non-White students, as is

widely expected but empirically unclear, they may contribute to the reproduction of social

inequality.

Inclusive Versus Exclusive Security

In Visions of Social Control (1985), Stanley Cohen distinguishes between exclusionary and inclu-

sionary efforts to regulate deviance that incorporates both crime prevention and punishment prac-

tices. These modes are distinguished by how they regard the deviant, their intervention strategies

and purposes, and likely consequences. Exclusionary approaches are more explicit and punitive,

directed at those perceived to be hard-core, intractable deviants, and intend to diminish or eliminate

social ties. Inclusionary approaches are more diffuse, benign, and therapeutic, intended for those

perceived to be near-normal or marginally deviant, and to reabsorb by maintaining and even

strengthening social ties. Although severe, formal and stigmatizing sanctions of capital punishment,

imprisonment, and physical restraint represent an ethos of exclusion, the inclusive ethos manifests in

less visible and less stigmatizing controls such as ‘‘discreet surveillance, data banks, [and] crime pre-

vention through environmental design’’ (Cohen, 1985, p. 221). Cohen anticipates a future of deeply

bifurcated or double-edged social control: a soft edge of ‘‘indefinite inclusion’’ or societal absorp-

tion maintained by therapeutic and technological controls over personal conduct and social space,

contrasting with a hard edge of ‘‘rigid exclusion,’’ maintained through the punitive classification and

control of outcasts (Cohen, 1985, pp. 232–233; also see Foucault, 1977; Young, 1999).

The Exclusionary Ethos and U.S. School Security. Hirschfield’s recent research has explored this ‘‘ethos

of inclusion versus exclusion’’ in contemporary school security. By disentangling the mix of school

security measures that might at first appear to be very similar, we can appreciate how even subtle

distinctions in measures have different social meanings or consequences, and these distinctions are

important to assessing the potential disintegrative effect of inequality in school security. Like

Cohen, he ties the inclusive/exclusive distinction to the function of the control measure as much

as to its particular form. Most importantly, inclusive security measures function to ‘‘reduce or

obscure differences among subjects,’’ while exclusionary school security measures function to

‘‘magnify deviant subjects for the purpose of exploiting or purging them’’ (p. 45). Inclusionary secu-

rity measures—such as surveillance cameras—involve an often imperceptible yet constant form of

discipline, ‘‘encourag[ing] students to direct their bodies toward normal, orderly, and productive




action,’’ without singling out particular students as deviants requiring control. By contrast, exclu-

sionary security measures, such as metal detectors and drug-sniffing dogs, more clearly communi-

cate and facilitate interests in identifying and removing deviant (rule-breaking or trouble-making)

students. Principals with an inclusionary ethos are ‘‘natural foes of metal detectors,’’ since the tech-

nology embodies exclusion, treating students with criminal suspicion, and erecting physical and pos-

sible legal barriers between students and learning environments (Hirschfield, 2010, p. 46).

Within the broad classifications of exclusionary and inclusionary measures, there are varieties of

practice or application that may enhance or diminish this function as a mechanism of inclusion or

exclusion. For example, although Hirschfield (2010) characterizes metal detectors as ‘‘inherently

anti-inclusionary,’’ he notes that the degree to which they facilitate magnification and exclusion

of deviant subjects depends somewhat on their use, including whether they are operated by school

personnel or police officers in the school. ‘‘Since the police who operate metal detectors cannot be

disciplined by school officials and are encouraged to make and tally arrests and seizures of contra-

band,’’ he writes (p. 43), they not only escalate the likely severity of sanctions for misconduct (i.e., a

detected weapon) but have used the metal detector screening as an occasion to conduct more exten-

sive surveillance (e.g., frisks) to identify ‘‘problem students’’ and arrest them for uncooperative

behavior (e.g., stepping out of line or refusing to be searched). Similarly, while a drug-sniffing dog

is an explicit crime-detecting measure intended to identify and purge offenders, and in this sense

exclusive, this would be a less stigmatizing and exclusionary measure if utilized discreetly—for

instance, during class periods—than if on regular patrol, stationed prominently in the cafeteria, or

alongside a metal detector.

To be sure, every security measure has the potential to be employed with elements of an exclusive

or inclusive ethos and is ambiguous in this sense. However, there is variable ambiguity. Some security

measures, such as locked gates and on-campus police, are highly ambiguous with respect to their

exclusive or inclusive function. Locked gates prevent movement in and out of school environments,

and campus protection and control purposes are difficult to disentangle. Campus security officers and

police are similarly ambiguous (Hirschfield, 2010). Although surveillance cameras, locked gates, and

even school police can be assimilated into the culture of the school, and employed to prioritize disci-

pline and inclusion rather than classification and exclusion, the metal detector and drug-sniffing dog

are less ambiguous interventions, clearly intended to detect criminal offenses and offenders. The

clearer exclusive functions of these measures (i.e., weapon and drug crime detection) provide a basis

for empirically assessing differential exposure to potentially disintegrative school security.

Competing Views of How Race and Poverty Relate to Exclusive Security

Our primary interest is to address the debate over whether the rise in exclusive school security is

generalized, reflective of a broad ‘‘law and order governmentality,’’ or concentrated in schools ser-

ving poor, non-White, or otherwise marginalized populations. We explain each of these perspectives

subsequently.

Social Reproduction and School Security. On one hand, a social reproduction perspective on inequality

in education (Bourdieu & Passeron, 1990; Bowles & Gintis, 1976) leads us to expect greater expo-

sure of poor and non-White students to exclusive security. Bourdieu and Passeron (1990) have

argued that schools reinforce the hierarchy of cultural capital that delineates among students.

Schools teach and reward middle-class norms while negatively appraising styles and mannerisms

that differ. This practice rewards middle-class students and punishes others by conflating cultural

capital with academic ability, thus reproducing and rationalizing existing social inequality by mak-

ing divisions among groups appear to be the product only of hard work and intelligence (for a thor-

ough review see Lewis, 2006). To the extent that youth with less cultural capital are perceived as




security threats, with more exclusive security directed their way, a social reproduction perspective

may help explain differential exposure to an inclusionary ethos of school security and its potential

consequences.

Consistent with this perspective, Hirschfield’s research on the inclusionary and exclusionary

ethos of school security makes frequent reference to racial and socioeconomic inequality. ‘‘While

suburban schools are hardly immune from criminalization,’’ Hirschfield (2008, p. 84) writes, ‘‘crim-

inalization in these contexts takes on more diluted or hybridized forms owing to the primacy of com-

peting ideals like consumer choice and individual freedom.’’ He partially attributes the exclusive

security faced by poorer and non-White youth to unique pressures of underresourced, underperform-

ing, and overcrowded schools.

Loic Wacquant’s description of exclusive security in public schools in the ‘‘hyperghetto’’ simi-

larly contends that select schools reproduce marginality by emulating prisons and treating margin-

alized youth as criminals, habituating their acceptance of ‘‘custody and control’’:

Public schools in the hyperghetto have . . . deteriorated to the point where they operate in the manner of

institutions of confinement whose primary mission is not to educate but to ensure ‘‘custody and con-

trol.’’ . . . The carceral atmosphere of schools and the constant presence of armed guards in uniform in the

lobbies, corridors, cafeterias, and playgrounds of their establishment habituates the children of the hyper-

ghetto to the demeanor, tactics, and interactive style of the correctional officers many of them are bound to

encounter shortly after their school days are over. (Wacquant, 2001, pp. 94–95, italics in original)

Here, marginalized youth are presumed to be young criminals and treated as such through expo-

sure to the hard edge of exclusive practices (e.g., police surveillance and metal detectors), while

youth with social, political, and cultural capital are presumed to be near normal and habituated for

social absorption in their selective exposure to inclusive security regimes.

Whatever the underlying social, economic, and political sources be, disparity in exposure to

exclusive school security can have profound consequences on students’ social mobility, since sus-

pension, expulsion, and arrest limit future educational, employment, and other prospects (e.g.,

Davies & Tanner, 2003; Fabelo et al., 2011; Hjalmarsson, 2008; Sweeten, 2006; Western, 2006).

Existing studies of differential exposure typically rely on limited data to establish these claims.

Although bivariate statistics do illustrate some of these relationships (such as the greater likelihood

of surveillance cameras at schools in suburbs and towns; see Robers et al., 2012, table 20.2), more

rigorous empirical analyses are needed to understand the extent of racial and socioeconomic dispar-

ity in exposure to exclusive security.

Governing Everyone Through Crime. In contrast, the relatively few studies looking at security measures

across schools have found inconsistent and limited evidence of social reproduction through the use

of exclusive security. In recent studies of school discipline and security in four high schools located

in two different states, for example, Kupchik (2009, 2010) has questioned the applicability of the

social reproduction thesis to comparisons between schools. As we discussed at the outset, he finds

that though socioeconomic and racial/ethnic compositions of student bodies significantly shape how

security measures are employed, the high schools he studied serving lower income youth of color

and those serving middle-class White youth have all adopted similar harsh, exclusive discipline and

security strategies that mirror criminal justice practices. As a result, he argues that there are more

similarities in these schools’ security strategies than the social reproduction perspective suggests.

This finding relates a central point in Jonathan Simon’s (2007) recent book, Governing Through

Crime, that the logic of crime control has become a dominating paradigm used to govern insecurity

and risk. One can see the influence of governing through crime in nationwide policies, such as a

requirement of zero-tolerance policies as well as financial incentives to schools for police and other




security measures, such as those provided in the ‘‘Secure Our Schools’’ Act of 2000 (see U.S. Con-

gress 2000; U.S. Department of Justice 2011). Simon notes that governing through crime has influ-

enced rules and practices in schools (and elsewhere) across social strata, as ‘‘The very real violence

of a few schools concentrated in zones of hardened poverty and social disadvantage has provided a

‘truth’ of school crime that circulates across whole school systems’’ (Simon, 2007: 210; see also

Lyons & Drew, 2006). Thus, this perspective leads us to predict that student body race and poverty

will have little effect on the distribution of security mechanisms, since exclusionary security has

spread throughout schools of all social strata. Given the lack of empirical research on national,

school-level differences in school security, it remains difficult to adjudicate between the differential

exposure (e.g., social reproduction) and more generalized ‘‘governing through crime’’ arguments;

our analyses offer an initial test of these competing theoretical perspectives.

School Security Environments: Conceptual and Grade-Level Distinctions. As we stated earlier, our analyses

differ from the few prior studies to consider school-level variation in school discipline and security

in two important ways. First, although the work of Payne and Welch (2010) and Welch & Payne

(2010, 2012) contribute to our understanding of how race and poverty relate to school punishment,

they do not consider security practices within schools. Irwin et al. (2013) add this piece to the lit-

erature, but in a fairly broad way, using summary measures of security guards (i.e., their measure

of personnel includes both security guards and police, and their measure of surveillance is a count

summing the use of varying technologies and practices) rather than examination of specific security

practices. There is a crucial distinction between punishment practices (e.g., suspension or expul-

sion), which are typically policy or otherwise normative responses to specific incidents, and how

built environments vary according to school security measures in place. The distinction between

security and punishment are often blurred in school security research, as scholars assess the overall

security and discipline climate of schools (e.g., Kupchik, 2010; Lyons & Drew, 2006; Nolan, 2011)

at the expense of analytical clarity. Certainly, both are important but distinct contexts of inclusionary

or exclusionary control. Here, we are particularly interested in the notion that all students—whether

subject to punishment or not—can be adversely affected by an ethos of exclusion, represented by the

normal presence of exclusive school security measures.

Second, we contribute to existing research by considering variation in security measures among

elementary, middle, and high schools. Prior research finds that the differential treatment of students

along socioeconomic and racial lines can begin early, at the very start of elementary school (see

Kozol, 1991; Rist, 1973; Weis, 1992), yet the literature on school security focuses mostly on high

schools (e.g., Casella, 2001; Lyons & Drew, 2006; Nolan, 2011) or occasionally on middle schools

as well (e.g., Ferguson, 2000; Payne & Welch, 2010; Welch & Payne, 2010, 2012), and in one case

on all levels of school but without distinguishing between them (Irwin et al., 2013). It is important

that we extend our analyses to consider different educational levels, as the relationship between race,

poverty, and exclusive school security may differ across school level. This could be the case if older

youth are viewed as more dangerous than younger students, or if the nationalization and federaliza-

tion of exclusive security has had distinct consequences for high school environments, making

exclusive security common to U.S. high schools, irrespective of school and student characteristics.

If exclusive security is more selectively employed at lower grade levels, racialized and classist con-

structions of danger or budget pressures in poor districts might shape these decisions, making race

and poverty more predictive of exposure for younger students.

Methods

To consider the variation in school security practices across schools, we analyze the restricted ver-

sion of the 2005–2006 School Survey on Crime and Safety (SSOCS). The SSOCS is a nationally




representative survey of school administrators managed by the National Center for Education

Statistics on behalf of the Federal Department of Education’s Institute of Education Statistics and

conducted by the U.S. Census Bureau. The 2005–2006 survey was the third such survey distributed

since 1999–2000. Respondents were selected using a stratified sampling design, with a sampling

frame of all regular public schools (and charter schools) from the National Center for Education

Statistics’ Common Core of Data Public School universe file, resulting in a sample of 2,7201

schools. In Table 1, we describe the characteristics of this sample; for additional details about the

sampling process, see Bauer, Tang, Derby, Parmer, and Swaim (2007).

Dependent Variables

The SSOCS data are well suited for our research question because they include a series of questions

about the use of both inclusionary and exclusionary security measures. We focus in particular on two

security measures that we described earlier as exclusionary: drug-sniffing police dogs and metal

Table 1. Summary Statistics.

Mean Std. Dev. Range

RegionMidwest 0.26 0.44 0–1South 0.24 0.43 0–1West 0.32 0.47 0–1Northeast 0.18 0.39 0–1

School characteristicsNontraditional school 0.07 0.25 0–1Number of students (ln) 6.47 0.79 1.39–8.54Neighborhood crime 1.30 0.58 1–3Location: urban 0.26 0.44 0–1Location: urban fringe 0.38 0.49 0–1Location: town 0.10 0.30 0–1Location: rural 0.26 0.44 0–1Parent involvement in school 2.55 0.77 1–4Disorder 4.03 0.55 1.5–5Violence (ln) 2.17 1.25 0–5.97Thefts (ln) 1.26 1.14 0–4.39Weapons offenses (ln) 0.77 0.77 0–3.43Alcohol/drugs (ln) 1.02 1.14 0–4.44Vandalisms (ln) 1.02 0.97 0–4.11Threats (ln) 1.49 1.35 0–5.86Attendance (ln) 4.54 0.20 0.69–4.62Percent ESL (ln) 1.40 1.23 0–4.62Percent special education (ln) 2.54 0.58 0–4.62Percent low test takers (ln) 2.30 0.94 0–4.62Percent racial/ethnic minority (ln) 3.03 1.21 0–4.62Percent free/reduced lunch (ln) 3.45 0.94 0–4.62

Dependent variablesDrug–sniffing dogs 0.38 0.49 0–1Metal detectors 0.09 0.28 0–1Police officer 0.38 0.49 0–1Surveillance cameras 0.55 0.50 0–1

Note. Std. dev ¼ standard devaiation.




detectors; one that we described earlier as inclusionary: surveillance cameras; and one that we

described earlier as fairly ambiguous: SROs. We consider the likelihood that the responding school

administrator answered affirmatively to questions about whether each of these four security mea-

sures is used in his or her school. Specifically, we constructed dichotomous variables to represent

the use of random checks using drug-sniffing dogs, whether either students or visitors must pass

through metal detectors (including random metal detector checks on students), the use of surveil-

lance cameras, and whether there is a full-time school resource officer or law enforcement officer.2

Independent Variables

In order to test whether schools serving racial/ethnic minority and lower income youth are more

likely to use exclusionary security mechanisms and less likely to use inclusionary security mechan-

isms, our primary independent variables are the racial/ethnic minority and low-income composition

of schools’ student bodies, as reported by the school administrator. To measure race/ethnicity, we

include the percentage of the enrolled students who are racial/ethnic minorities. Our primary mea-

sure of poverty comes from the percentage of students who receive free or reduced-price lunch.3

As controls, we also include variables for the percentage of students who have limited proficiency

in English, who are enrolled in special education curriculum, and who score below the 15th percen-

tile on standardized tests, since each may be an alternate measure of marginalization or subordinate

status (see Noguera, 2003). English proficiency is likely related to immigration status and may

denote a lack of political capital (Levine-Rasky, 2009), while large numbers of poor standardized

test performers are a liability for schools in an era of accountability via high-stakes testing (see

Lawrence, 2006; Simmons, 2007).4

We include a variable indicating parental involvement in school academics and social events, since

it both relates to student body socioeconomic status (Lareau, 2003) and has the potential to shape

school security practices (Kupchik, 2010; Noguera, 2003). This variable is the mean response to a

series of questions including the percentage of parents who participate in open house or back to school

night, participate in teacher/parent conferences, participate in subject area events, and volunteer at the

school. Each of these original variables is coded along a 4-point scale (1�25%, 2¼ 26–50%, 3¼ 52–

75%, 4¼ 76–100%), and the resulting index achieved high interitem reliability (Cronbach’s a ¼ .82).

We control for the extent to which school security varies across geographic regions. To the extent

regions maintain distinct ‘‘cultures of control’’ (Barker, 2009; Lynch, 2009), school security may

vary according to the regional locations of schools. Although neither individual schools nor states

are identifiable in the SSOCS, the data set includes a variable for region. Using this, we include

dummy variables for midwest, south, and west regions, which allow us to compare school security

in these regions relative to schools in the northeastern United States.

Another important potential predictor of the school security environment is the actual problem of

school misbehavior as well as crime and delinquency in the school and surrounding community

(Devine, 1996; Nolan, 2011). To consider whether schools with greater crime problems might

implement varying security measures to deal with these problems, we include a number of variables

related to levels of student and school area crime and disorder. Each of these variables measures the

total number of incidents of a certain type of misbehavior or crime: violence (including the total

number of rapes, sexual batteries, robberies with weapon, robberies without weapon, attacks with

weapon, and attacks without weapon), weapons (number of possession of firearms and possession

of knives/sharp objects), alcohol and drugs (number of possession or use of alcohol and distribution

of drugs), and threats (number of threats with a weapon and threats of attack without weapon). We

also include the reported number of incidents of theft/larceny and vandalism. Additionally, we com-

puted an index measuring school disorder; this index is the mean of responses to a series of questions

about the perceived frequency (on a scale of 1–5 that was recoded so that higher values now reflect




greater frequency) of the following problems: student racial tensions, student bullying, student sex-

ual harassment of student, verbal abuse of teachers, student disorder in classroom, student acts of

disrespect, student gang activities, and student cult or extremist activities (Cronbach’s a¼ .81). Fur-

ther, we control for administrators’ perceptions of the level of crime in the neighborhood surround-

ing each school, along a scale of 1 to 3, recoded so that greater values reflect higher perceived levels

of crime. As a final control for perception of student behaviors, we also include the reported average

daily attendance rate.

Importantly, we recognize the inability of these student behavior variables to capture a causal

mechanism. Since these data are cross-sectional, we are unable to interpret whether reported inci-

dence of violence, for example, are objective indicators of behavior or the result of school sensitivity

to these issues. It is likely that school concerns and policies related to security are incident generat-

ing themselves, somewhat irrespective of actual student or community youth misbehavior. Rather

than assessing causal mechanisms, these variables offer two contributions to our analyses: (1) as

control variables, they help isolate how student social status is related to security practices net of

actual or perceived misbehavior and crime rates and (2) if significant, they will suggest the need for

additional research to consider a causal mechanisms using data more properly suited to the task.

Other control variables include the size of the student population, whether the school is a nontradi-

tional (magnet or charter) public school, and location (we include dummy variables for urban fringe,

town, and rural, with urban excluded as the contrast). School population size is relevant here because it

may shape levels of student victimization (see Gottfredson & DiPietro, 2011) and thus indirectly influ-

ence security measures. We control for nontraditional schools because of their demonstrated organiza-

tional differences in how they approach their educational mission (see Renzulli, Parrott, & Beattie,

2011), and we control for location in order to control for the intense level of security that has been

documented in urban public schools (e.g., Mukherjee, 2007; Nolan, 2011).

Several of our continuous independent variables, such as student population and percentage of

students in special education curriculum, are positively skewed and introduce the potential problem

of influential outliers. To reduce this potential problem, we use the natural logarithm of each con-

tinuous variable5; we do this for all independent variables other than the dummy variables (region,

location, and nontraditional school), our one ordinal variable (crime at school), and the indexes com-

puted as means (parental involvement and school disorder).

Analytic Strategy

Because each of our four dependent variables is dichotomous, we use logistic regression to estimate

the odds of a school using each security practice. Given the many differences one might find across

different levels of schooling, we consider elementary schools, middle schools, and high schools sep-

arately; we exclude schools that identify as being combined grade levels (n¼ 140). After estimating

our primary models, we then further explore these results by estimating the predicted probability of

different categories of schools using each security practice; here, we manipulate the percentage of

students who receive free or reduced-price lunch while keeping all other variables at their means.

All analyses are performed using survey-specific commands in Stata/SE 11.1. The models are

adjusted in two ways. One is that we use the weighting variable provided in the data set to accom-

modate for unequal probability of selection in the survey sample; the provided weight handicaps

each case’s score to reflect the probability of sample selection (see Bauer, Tang, Derby, Parmer,

& Swaim, 2007). The second is that since we compute models separately for each level of school,

we specify the subpopulation included in each analysis. This option within Stata accommodates sub-

populations while still accurately computing standard errors; without specifying the subpopulation,

the standard errors might be biased, since the survey design structure assumed within the Stata sur-

vey command would no longer be valid.




Prior to releasing the data set, the National Center for Education Statistics imputes missing values

for most variables (for details, see Bauer et al., 2007). The only variable in our analyses for which

missing data are not imputed is the percentage of racial/ethnic minority students in each school. We

hesitate to impute missing values for this, since it is central to our analysis and we do not wish to

distort the data in any way and also because missing data on this variable represent a relatively lim-

ited problem. Across school levels, there are a total of 70 cases with missing data on this variable,

which we exclude from the analysis. After excluding these cases and the combined grade-level

schools, our sample includes n¼ 700 elementary schools, n¼ 920 middle schools, and n¼ 890 high

schools (total sample n ¼ 2,510).6

Results

Logistic Regression Models

The results of the logistic regression models are presented in Tables 2 and 3; Table 2 shows results

for exclusionary security measures (metal detectors and drug-sniffing dogs) and Table 3 for inclu-

sionary security (surveillance cameras) and ambiguous security (police officers). Each table lists the

exponentiated coefficients, Exp(B), or the difference in the odds of a school having each security

mechanism associated with an increase of one in each independent variable, separately for each

school grade level. As we described earlier, we are particularly interested in how school-level

race/ethnicity and socioeconomic status shape the likelihood of schools relying on exclusionary

(drug-sniffing dogs, metal detectors), inclusionary (surveillance cameras), and ambiguous (police)

security measures. These results are highlighted using boldface type: significant positive results for

the exclusionary measures would support the social reproduction perspective of discrepancies across

race and class groups, while nonsignificant results would support the ‘‘governing everyone through

crime’’ perspective.

Our results for student body race/ethnicity illustrate the importance of considering distinct secu-

rity measures separately, since we find a consistent and significant effect for the presence of metal

detectors but not for other security practice. On one hand, regarding most school security measures,

we see that schools with various populations of minority youth have implemented similarly exten-

sive security regimes. This finding suggests some convergence, in that extensive school security sys-

tems are found across social strata, and diverse students have grown subject to similar forms of

control (see Kupchik, 2010; Simon, 2007). On the other hand, schools with larger proportions of

racial/ethnic minorities are significantly more likely than others to have metal detectors, an exclu-

sionary security mechanism. Importantly, this is true for each level of school (elementary, middle,

and high) and after controlling for both locations in an urban setting and area crime rates. Since we

also control for weapons offenses at schools, this result seems to be driven by assumptions of threat

rather than observations, that is, school officials may respond to often implicit associations between

racial/ethnic minorities, violence, and use of weapons (Eberhardt, Goff, Purdie, & Davies, 2004).

We do find results that contradict a social reproduction perspective as well, though they are less

consistent or robust than the relationship between race/ethnicity and metal detectors. High schools

with larger proportions of minorities are less likely to use drug-sniffing dogs, and middle schools

with larger non-White populations are more likely to have surveillance cameras (an inclusionary

measure), though neither of these effects is significant at other school levels.

Our measure of poverty, the percentage of students receiving free/reduced lunch, is significant

and positive in 6 of the 12 models. Higher percentages of poor students are predictive of greater odds

that schools have drug-sniffing dogs in each school level, supporting the hypothesis that students

serving poorer youth are more likely to rely on exclusionary security. The percentage of poor stu-

dents also relates positively to the odds of metal detectors in middle schools, again in support of this




perspective. And, we find that police, a measure we categorize as neither inclusionary nor exclusion-

ary, are more likely in both elementary and middle schools with more poor youth. These findings

suggest that general trends toward policing in schools and criminalizing student misconduct may

be unique structural features of schools serving the poorest youth and communities.

Importantly, our results suggest that the effect of socioeconomic status is greatest and most con-

sistently found early in the educational process, since the effect of poverty status is most consistently

observed at the elementary or middle school levels rather than in high schools. When it comes to

young children, poverty appears to have particular salience as a marker of the need for exclusionary

school security and a law enforcement presence. This also suggests that the experience of policing in

school and criminalization of student misconduct begins earlier for students attending schools with

concentrated poverty, potentially contributing to short- and long-term disparities in educational

achievement. We interpret this result to suggest that in elementary and middle schools, where youth

are less fearsome as a group (compared to high schoolers) and most security measures are less com-

monly found than in high schools, poverty is taken to indicate dangerousness. Resulting differential

exposure to exclusionary security is an added and perhaps compounding component of the concen-

trated disadvantage in these schools.

Table 2. Logistic Regression Models Predicting Odds of Exclusionary Security Practices, by School Level(Exp(B)).

Metal Detectors Drug-Sniffing Dogs

Elem. Middle High Elem. Middle High

RegionMidwest 78.570*** 1.296 0.399* 3.932 3.364*** 3.670***South 11.920 1.747 1.513 0.193 5.467*** 6.496***West 4.685 1.387 0.355* 1.921 4.552*** 4.562***

Nontraditional school 0.497 1.816 0.836 0.548 0.632 0.369**Students (ln) 0.814 6.761*** 1.517 1.685 1.428 1.750**Location

Urban fringe 0.724 0.495* 0.369** 13.82* 1.411 1.682*Town 1.454 0.697 0.340* 18.26* 1.892* 2.669**Rural 0.741 0.905 0.301** 83.990*** 2.790*** 2.233**

Parent involvement 0.242* 0.667 1.108 1.120 0.854 1.042Neighborhood crime 1.109 1.910** 1.790* 0.494 0.944 0.456***Crime/misbehavior

Disorder (ln) 10.54* 1.971** 1.690* 0.324 0.895 0.952Violence (ln) 1.215 0.945 0.877 0.659 0.874 0.812Thefts (ln) 1.700 0.889 1.039 2.891* 1.140 0.962Weapons (ln) 2.556* 1.061 1.312 0.842 0.642*** 0.954Alcohol/drugs (ln) 5.231* 0.943 1.078 6.810* 1.420** 0.984Vandalisms (ln) 2.004 0.968 1.210 0.344* 0.924 1.178Threats (ln) 0.878 0.981 0.815 0.729 1.030 0.972

Avg. daily attendance (ln) 1.369 1.315 1.750 4.654 129.200 1.079% ESL (ln) 0.791 0.619*** 0.664** 0.788 1.042 0.806*% Special education (ln) 0.621 1.413 1.220 0.688 1.040 1.367% Low test scores (ln) 0.767 0.963 1.166 1.135 0.898 0.944% Racial/ethnic minority (ln) 10.420* 2.309** 1.814** 1.554 0.866 0.719**% Free/reduced lunch (ln) 0.418 2.039* 1.717 4.401** 1.374* 1.719***F 3.891 6.485 6.339 3.527 5.339 6.237

Note. Elem ¼ elementary.*p < .05. **p < .01. ***p < .001.




Results for our measure of parental involvement are interesting as well. High schools with higher

levels of parental involvement in academic and social events (open houses, parent/teacher confer-

ences, subject area events, and volunteer efforts) are more likely to have a police officer. However,

elementary schools with higher levels of parental involvement in academic and social events are less

likely to have metal detectors. Assuming that greater parental involvement provides some parental

capacity to shape school policy (see Blankenau & Leeper, 2003), this result suggests that parents

may advocate for police officers in schools as protection for their children (see Casella, 2001) while

resisting the more exclusionary use of metal detectors.7

The percentage of youth who are in English as a second language class is negatively related to the

odds of having cameras in middle schools, metal detectors in middle school and high school, and

drug-sniffing dogs in high schools. The fact that this variable predicts reduced odds of having cameras,

metal detectors, and drug-sniffing dogs suggests that schools with large concentrations of immigrants

are less likely than others to implement both inclusionary and exclusionary security measures. As we

discuss subsequently, additional data that allows us to disentangle different minority groups would allow

us to explore this result in greater depth. The percentage of students who are in special education classes

and the percentage who score low on standardized tests are weak predictors of school security measures.

Table 3. Logistic Regression Models Predicting Odds of Inclusionary and Ambiguous Security Practices, bySchool Level (Exp(B)).

Inclusionary: Surveillance Cameras Ambiguous: Police Officers

Elem. Middle High Elem. Middle High

RegionMidwest 0.792 0.950 1.370 0.375 1.079 0.966South 0.553* 0.761 1.453 0.612 2.327** 4.624***West 0.218*** 0.503** 1.107 0.321* 0.996 1.327

Nontraditional school 1.388 0.809 0.509 1.211 0.868 1.698Students (ln) 1.516* 1.386 2.349*** 1.595 3.379*** 2.804***Location

Urban fringe 1.091 1.432 1.516 1.246 0.497*** 0.780Town 1.061 1.548 1.655 0.896 0.622 1.084Rural 0.739 1.738* 1.295 1.042 0.712 0.946

Parent involvement 0.791 0.861 0.988 1.857 1.106 1.434*Neighborhood crime 1.288 0.987 1.175 0.752 0.929 1.317Crime/Misbehavior

Disorder (ln) 0.813 1.312 0.693 0.690 1.058 0.938Violence (ln) 0.956 1.037 1.124 0.936 1.087 1.030Thefts (ln) 1.048 0.999 1.004 1.132 1.134 1.255*Weapons (ln) 1.448* 0.999 1.017 2.072** 0.829 1.487*Alcohol/drugs (ln) 0.582 1.178 0.738* 0.444 1.131 1.103Vandalisms (ln) 0.840 0.958 0.820 0.821 0.904 1.070Threats (ln) 0.850 1.007 0.955 0.881 0.983 0.929

Avg. daily attendance (ln) 1.179 0.805 1.010 1.548 0.450 1.021% ESL (ln) 0.992 0.782** 0.932 1.288 1.094 1.029% Special education (ln) 0.992 1.300 1.115 1.640 1.102 1.057% Low test scores (ln) 0.962 1.067 1.160 0.879 1.117 1.014% Racial/ethnic minority (ln) 1.073 1.247* 1.055 0.934 1.173 1.137% Free/reduced lunch (ln) 0.924 1.092 1.208 2.255** 1.408* 1.096F 2.527 2.049 3.008 2.393 6.182 8.469

Note. Elem ¼ elementary.*p < .05. **p < .01. ***p < .001.




School size and location also relate to security measures. Larger schools are more likely to have each

of the security mechanisms among at least one school level, presumably because larger schools can be

more chaotic and difficult to manage. Region is an important consideration in criminal punishments,

broadly, and it appears to be an important but inconsistent determinant of school security as well. Schools

in each region other than the northeast—and especially in the south—are more likely than those in the

northeast to use police officers and drug-sniffing dogs. It appears that school security mirrors broader

punishment and social control trends, with more exclusionary and punitive practices in the south (e.g.

Lynch, 2009). Urbanicity matters as well, even while controlling for race/ethnicity and socioeconomic

status of student bodies. Urban high schools are more likely than high schools in urban fringe areas,

towns, and rural areas to have metal detectors; in conjunction with the robust results linking race/ethni-

city to the presence of metal detectors, this result suggests that urban high schools hosting large popula-

tions of youth of color are distinguished by the presence of metal detectors but not other forms of school

security. Yet urban schools are less likely than other schools at each level to have drug-sniffing dogs, our

other measure of exclusionary security, with rural schools especially likely to use this security measure.

Average daily attendance is not significantly related to any security measure, and the various

measures of crime and misbehavior are only weakly and inconsistently related to school security.

Perceived crime in the school neighborhood is positively related to metal detectors in middle and

high schools and negatively related to drug-sniffing dogs in high schools. Incidents of student vio-

lence do not significantly relate to any school security measure. Weapon incidents are related to

increased odds of surveillance cameras in elementary schools, police in elementary and high

schools, and metal detectors in elementary schools but decreased odds of drug-sniffing dogs in mid-

dle schools. Alcohol and drug use is a significant predictor of the use metal detectors in elementary

schools, of drug-sniffing dogs in elementary and middle schools, and of reduced odds of having sur-

veillance cameras in high school. Finally, our disorder scale, which consists of problems such as

student racial tensions, bullying sexual harassment, and others, significantly predicts increased odds

of the presence of metal detectors at each school level. Although we are unable to test causal rela-

tionships with these variables, due to the problem of specifying temporal order, we nonetheless find

it interesting that there are few consistent relationships between reported levels of students’ criminal

behaviors and the adoption of school security measures. These results clearly mirror prior sugges-

tions that school security is a response to perceived, not actual, threats in school (see Cornell, 2006).

Predictions for School Categories of Free/Reduced-Price Lunch

Our primary independent variable related to poverty, the percentage of students receiving free/

reduced-price lunch, was found to have a consistent and robust effect on various school security

practices, particularly among elementary and middle schools. To better illustrate the effect of pov-

erty, we predict the probability of different categories of elementary and middle schools having each

security mechanism. While holding all other variables at their means, we estimate the probability of

elementary and middle schools having each security practice under three conditions: (a) concen-

trated advantage: 10% receiving free or reduced-price lunch, (b) integration: 50% receiving free

or reduced-price lunch, and (c) concentrated disadvantage: 90% receiving free or reduced-price

lunch. These predicted probabilities are graphed in Figure 1 for elementary schools and in Figure

2 for middle schools. They show a fairly consistent pattern whereby security measures, particularly

exclusionary security, are more likely in conditions of concentrated disadvantage. This pattern is

clear regarding drug-sniffing dogs in both school levels, and metal detectors in middle schools,

though there are too few metal detectors in elementary schools overall (2.5%) for any pattern to

be clear. We also see that surveillance cameras, an inclusionary measure, are more common in ele-

mentary schools of concentrated advantage, while officers are more common in schools with con-

centrated disadvantage.




Figures 1 and 2 illustrate the importance of considering cumulative effects on the link between

student characteristics and types of school security measures. If one were to consider only regression

coefficients for student socioeconomic status, as prior research tends to do, then important substan-

tive effects of concentrated advantage and disadvantage might go overlooked. Instead, by distin-

guishing schools according to thresholds of socioeconomic stratification, we identify an apparent

buffering effect of concentrated advantage, and the preponderance of exclusionary security mea-

sures in elementary and middle schools of concentrated poverty.

Discussion and Conclusions

By analyzing nationally representative data from public schools, we provide an empirical examina-

tion of how school and student body characteristics relate to the use of inclusionary and exclusionary

Figure 1. Estimated probabilities of elementary schools using security practices, by category of school.

Figure 2. Estimated probabilities of middle schools using security practices, by category of school.




school security measures. Our results generally support prior arguments linking race/ethnicity and

poverty to school-level punishment policies (see Irwin et al., 2013; Payne & Welch, 2010; Welch

& Payne, 2010, 2012) while extending this literature in important and new ways by considering par-

ticular security measures and differences across grade levels. Consistent with Cohen (1985) and

Hirschfield’s (2008, 2010) application of Cohen’s work to school security, we find that exclusionary

measures (metal detectors, drug-sniffing dogs) are more prevalent at schools with more racial/ethnic

minority and low-income students, while inclusionary social control is more common at more

advantaged schools (see also Irwin et al., 2013).

With regard to student body race/ethnicity, our results clearly demonstrate that the specific exclu-

sionary security measure of metal detectors is more common in schools serving large numbers of

youth of color. Consistent with the social reproduction perspective, this result suggests that

racial/ethnic minority youth are exposed at greater rates to a practice that seeks to identify offending

youth and divert them to the criminal justice system. With regard to poverty, we find that exclusion-

ary security measures are more common among elementary and middle schools but not high schools.

Thus, poverty seems to mark younger children as potential threats requiring exclusionary control

measures, and this differential exposure to disintegrative security appears to compound the concen-

trations of disadvantage in such schools.

Although these relationships are clear in our analyses, it is also clear that the alternate theoretical

framework, of governing everyone through crime, has some validity as well. Overall, there is no

consistent effect of student body race/ethnicity regarding security measures other than metal detec-

tors, and little effect of student poverty on security measures among high schools. This suggests that

other than the use of metal detectors, both inclusionary and exclusionary security mechanisms are

part of the social fabric of high schools across social strata. The fact that we find support for both

of these perspectives highlights the need to consider the context of school security rather than offer-

ing generalizations. As we stated earlier, most research on this topic considers only high schools and/

or fails to distinguish between distinct security measures, and as a result the complex relationships

between race/ethnicity, poverty, and school security are either overlooked or important distinctions

are muddled.

Given the importance of educational experiences in socializing youth to future social roles and as

a credentializing process, we are particularly concerned about how the distribution of school security

measures may contribute to existing social inequality. Consistent with social reproduction argu-

ments discussed earlier, our findings suggest that potentially ‘‘disintegrative’’ effects of exclusion-

ary school security are concentrated in schools serving more minority and poor students, which may

institutionalize racial and socioeconomic inequality in American education and, ultimately, Amer-

ican society. The observed relationship between concentrated poverty and security at earlier grade

levels is especially troubling, as it suggests that disintegrative effects of school security are distinctly

borne by already marginalized students at relatively young ages. The extent to which divergent secu-

rity practices impact the long-term social, academic, and professional outcomes of youth awaits

future research and will hopefully be considered in developing future school security policy.

Our findings not only support but also extend prior research in various ways while also raising

questions that await further research. Our school-level findings mirror individual-level research

on school discipline, suggesting that teachers and school administrators rely on racist and classist

stereotypes of threat in interpreting student behavior, while raising additional concerns about insti-

tutionalized race and socioeconomic inequality in American education. Unfortunately, our data offer

limited insight into the underlying bases of these developments. Although recent attitudinal and cog-

nitive research on the race and social class politics of public punitiveness (e.g., Pickett & Chiricos,

2012; Unnever & Cullen, 2010) may help explain the distribution of exclusive school security pol-

icies,8 without data on attitudinal measures we are unable to empirically assess how the adoption of

various school security measures is related to racial and socioeconomic animus generally.




The observed prominence of exclusive school security in schools with more poor students and

youth of color is also likely a reflection of a number of structural conditions, including the federa-

lization of crime control. Lisa Miller (2008) has argued that the federalization and nationalization of

crime control has elevated the crime control agenda of more influential constituencies while margin-

alizing the poor and non-Whites. Federal funding for school security may illustrate this influence,

and future research should examine the role of federal financial incentives in shaping the implemen-

tation of inclusionary versus exclusionary school security measures. Earlier, we refer to the Secure

Our Schools Act of 2000, which provided funding for measures such as metal detectors, locks, light-

ing, security assessments, and security training. It is unclear whether the availability of federal fund-

ing for these measures has encouraged financially unstable school districts to adopt these measures

in greater numbers than other schools who may have the means to resist such narrowly targeted

financial incentives. Since schools receive their funding through complex and varying combinations

of local, state, and federal sources, we are unable to consider funding source in our analyses. Yet we

recognize that this public funding is a clear illustration of governing through crime and suspect that it

has an important role in both the proliferation and the unequal distribution of school security

measures.

The current study also lays the groundwork for continuing investigation into the use of police

officers across varying school contexts. As we discussed earlier, police officers in schools can

take on both inclusion- and exclusion-focused roles. Despite their ambiguity, we include the use

of police officers here because they are a high-profile and common response to widespread anxieties

about school safety. Our results show that they are more commonly placed in elementary and middle

schools with higher percentages of low-income youth, though it is not clear exactly what this means,

or whether they assume similar functions in schools across social strata.9

In addition to extending analyses to consider long-term consequences of school security, future

research can also address a number of limitations to the current study. One such limitation, which we

discussed earlier, is our inability to analyze whether student crime and disorder has a causal role in

the implementation of security measures. The use of cross-sectional data may also obscure the extent

to which race/ethnicity or poverty may have complex, nonrecursive relationships with security prac-

tices; we are unable to detect, for example, whether White or middle-class families leave schools

after metal detectors or drug-sniffing dogs appear. Our data are also limited in detail on the appli-

cation of security measures, for example, how metal detectors and officers are actually employed,

which limits our precision in operationalizing the inclusionary/exclusionary distinction. Another

limitation is that we do not assess the impact of community-level factors, such as poverty, unem-

ployment, and the politics of school governance (e.g., local vs. central control), which may contrib-

ute to the extent to which students are perceived as dangerous (see Hirschfield, 2008). A number of

additional variables would benefit future studies, especially measures related to race and socioeco-

nomic stratification (e.g., non-White representation among school authorities; educational attain-

ment of parents) and more detailed race/ethnicity information, which would allow us to consider

different racial/ethnic groups distinctly (see Peguero, 2011) and would provide greater insight into

the relationship between race, poverty, and exclusionary school security. Furthermore, although we

set out to assess the extent of differential exposure to exclusive security, it is crucial that future

research examines the actual implications of this inequality, including their suspected disintegrative

effects.

A final limitation of our analyses concerns our interpretation of the pronounced effect of race and

poverty among elementary and middle schools. Earlier we interpret this result to suggest that race

and poverty are taken to be particularly salient markers of potential criminality among younger chil-

dren; but an alternate possibility is that the poor youth and youth of color who receive the most atten-

tion from school disciplinarians drop out or are expelled before reaching high school. As a result,

perhaps the more modest results regarding high schools are due to a selection process instead.




We are now decades into a national, federally subsidized investment in school security measures.

Contemporary schools evidence elaborate security systems, somewhat irrespective of region, stu-

dent status characteristics, and level of education, consistent with this historical development. Our

findings confirm while also specifying previous suggestions that race and socioeconomic inequal-

ities relate to the use of exclusionary security measures. Although further research is needed to con-

firm whether or not and how such disparities in school security translate into the reproduction of

inequality, the prominence of security systems across high schools in particular suggests some con-

vergence rather than bifurcation (Cohen, 1985) in the experience of social control and raises a more

general social reproduction question. That is, how has this era of unprecedented policing, surveil-

lance, and other crime control strategies within American schools socialized this generation of stu-

dents? How does this new school security environment affect school safety and, as significantly,

educational achievement or aspiration? Has criminal justice oriented security rendered schooling

a disintegrative experience? These questions are important to consider for all students but perhaps

most of all those whose formative early educational experiences are in elementary and middle

schools marked not only by concentrations of poverty but also by police officers, metal detectors,

and drug-sniffing dogs.

Acknowledgments

We wish to thank Kelly Welch for her comments on a previous draft of this manuscript and the anonymous

reviewers for their feedback and support.

Authors’ Note

An earlier version of this article was presented at the Annual Meetings of the American Sociological

Association.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publi-

cation of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

1. As required by the Institute of Education Statistics, all reports of sample sizes are rounded to the nearest 10

throughout this article.

2. The term ‘‘school resource officer’’ (SRO) is used to describe police officers who are stationed in schools.

SROs usually wear their full police uniforms and carry firearms as do their counterparts outside of schools

(see Kupchik, 2010). It is important to note that unlike Irwin et al. (2013), we consider only law enforcement

officers here rather than security guards to better focus on a security practice that has been a focus of policy

debate.

3. We realize this variable may not be completely accurate, if lower income families do not know about the

lunch program or refuse to apply for help with lunch purchases or if other families misrepresent their income

to receive reduced-price lunches. However, this is by far the best measure of socioeconomic status included

in the data, and it is consistent with prior research (e.g., Payne & Welch, 2010; Welch & Payne, 2010).

4. In earlier models, we introduced interaction terms of Percentage Non-White � Percentage Free/Reduced

lunch to capture the potential intersection of race and poverty in predicting school security but excluded

these from the final model because they did not improve model fit. We also introduced variables measuring

the square of percentage non-White and of percentage ree/reduced lunch to consider whether either variable




relates to school security in a curvilinear way, but excluded these from the final model as well because they

did not consistently improve the model fit.

5. We use the natural log of each continuous variable þ1, in order to remove values of 0 (which are undefined

when taking a logarithm).

6. Collinearity diagnostics suggested that multicollinearity is not a problem in our models; the average variance

inflation factor (VIF) is 1.69 and the highest value is 2.41.

7. To consider whether the impact of parents in particular on school security depends on student demographics,

in preliminary analyses we included interaction terms using our Parental Involvement Scale � Percentage

Minority and Parental Involvement � Percentage Receiving Free/Reduced Lunch. These interaction terms

did not improve the fit of our models, and thus we did not retain them.

8. Payne and Welch 2010 and Welch and Payne (2010, 2012) consider percentages of racial/ethnic minorities

to indicate racial threat, though unlike Pickett and Chiricos (2012) and Unnever and Cullen (2010), they do

not include measures of racial animus or other racialized attitudes.

9. In supplemental analyses, we find that among schools with a full-time police presence, the percentage of

youth receiving free/reduced-price lunch is positively related to the use of exclusionary security measures

(metal detectors, drug-sniffing dogs, or both), suggesting that police are used in conjunction with exclusion-

ary security in schools serving poor youth. Although we do not discuss these analyses here for the sake of

parsimony, they suggest fruitful future inquiries into the use of varying combinations of security measures.

References

Arum, R. 2003. Judging school discipline: The crisis of moral authority. Cambridge, MA: Harvard University

Press.

Arum, R., & Velez, M. (2012). Improving learning outcomes: School discipline and student achievement in

comparative perspective. Stanford, CA: Stanford University Press.

Ayers, W., Dohrn, B., & Ayers, R. (eds.). (2001). Zero tolerance: Resisting the drive for punishment in our

schools. New York, NY: New Press.

Barker, V. (2009). The politics of imprisonment: How the democratic process shapes the way America punishes

offenders. New York, NY: Oxford University Press.

Bauer, L., Tang, S., Derby, T. L., Parmer, R. J., & Swaim, N. L. (2007). School survey on crime and safety:

2005–06 data file user’s manual (NCES 2007-358). Washington, DC:National Center for Education Statis-

tics, Institute of Education Sciences, U.S. Department of Education.

Blankenau, J., & Leeper, M. (2003). Public school search policies and the ‘Politics of Sin’ Policy Studies Jour-

nal, 31, 565–584.

Blume, Howard. (2011). New policy aims to reduce truancy tickets in LAUSD. Los Angeles Times, October 21.

Retrieved July 17, 2013, from http://articles.latimes.com/2011/oct/21/local/la-me-1021-truancy-20111021

Bourdieu, P., & Passeron, J. (translated by Richard Nice). (1990). Reproduction in education, society, and cul-

ture (2nd ed.). Newbury Park, CA: Sage.

Bowditch, C. (1993). Getting rid of troublemakers: High school disciplinary procedures and the production of

dropouts. Social Problems, 41, 493–509.

Bowles, S., & Gintis, H. (1976). Schooling in capitalist America: Educational reform and the contradictions of

economic life. New York, NY: Basic Books.

Brady, K. P., Balmer, S., & Phenix, D. (2007). School-police partnership effectiveness in urban schools.’’ Edu-

cation and Urban Society, 39, 455–478.

Brooks, K., Schiraldi, V., & Ziedenberg, J. (2000). School house hype: Two years later. Washington, DC: Jus-

tice Policy Institute.

Burns, R., & Crawford, C. (1999). School shootings, the media, and public rear: Ingredients for a moral panic.

Crime, Law and Social Panic, 32, 147–168.

Casella, R. (2001). Being down: Challenging violence in urban schools. New York, NY: Teachers College

Press.



http://articles.latimes.com/2011/oct/21/local/la-me-1021-truancy-20111021


Casella, R. (2006). Selling us the fortress: The promotion of techno-security equipment for schools. New York,

NY: Routledge.

Cohen, S. (1985). Visions of social control: Crime, punishment and classification. Cambridge, MA: Polity

Press.

Cornell, D. G.(2006). School violence: Fears versus facts. Mahwah, NJ: Lawrence Erlbaum.

Davies, S., & Tanner, J. (2003). The long arm of the law: Effects of labeling on employment. The Sociological

Quarterly, 44, 385–404.

Devine, J. (1996). Maximum security: The culture of violence in inner-city schools. Chicago, IL: University of

Chicago Press

Downey, D. B., & Pribesh, S. (2004). When race matters: Teachers’ evaluations of students’ behavior. Sociology

of Education, 77, 267–282.

Eberhardt, J. L., Goff, P. A., Purdie, V. J., & Davies, P. G. (2004). Seeing black: Race, crime, and visual

processing. Journal of Personality and Social Psychology, 87, 876–893.

Eitle, T. M., & Eitle, D. J. (2004). Inequality, segregation, and the overrepresentation of African Americans in

school suspensions. Sociological Perspectives, 47, 269–287.

Fabelo, T., Thompson, M., Plotkin, M., Carmichal, D., Marchbanks, M., & Booth, E. (2011). Breaking

schools’ rules: A statewide study of how school discipline relates to students’ success and juvenile justice

involvement. New York, NY: Council of State Governments Justice Center and Public Policy Research

Institute.

Ferguson, A. A. (2000). Bad boys: Public schools and the making of black masculinity. Ann Arbor, MI:

University of Michigan Press.

Foucault, M. (transl. by Alan Sheridan). (1995). Discipline and punish: The birth of the prison (2nd ed.). New

York, NY: Vintage Books.

Gottfredson, D. C. (2001). Schools and delinquency. New York, NY: Cambridge University Press.

Gottfredson, D. C., & DiPietro, S.M. (2011). School size, social capital, and student victimization. Sociology of

Education, 84, 69–89.

Gottfredson, G. D., Gottfredson, D. C., Payne, A. A., & Gottfredson, N. C. (2005). School climate predictors of

school disorder: Results from a national study of delinquency prevention in schools. Journal of Research in

Crime and Delinquency, 42, 412–444.

Hirschfield, P. (2008). Preparing for prison: The criminalization of school discipline in the USA. Theoretical

Criminology, 12, 79–101.

Hirschfield, P. (2010). School surveillance in America: Disparate and unequal. In T. Monahan & R. D. Torres

(eds.), Schools under surveillance: Cultures of control in public education (pp. 38–54). New Brunswick, NJ:

Rutgers University Press.

Hjalmarsson, R. (2008). Criminal justice involvement and high school completion. Journal of Urban Economics,

63, 613–630.

Irwin, K., Davidson, J., & Hall-Sanchez, A. (2013) The race to punish in American schools: Class and race

predictors of punitive school-crime control. Critical Criminology, 21, 47–71.

Kim, C. Y., Losen, D. J., & Hewitt, D. T. (2010). The school-to-prison pipeline: Structuring legal reform. New

York: NYU Press.

Kozol, J. (1991). Savage inequalities: Children in America’s schools. New York, NY: Harper Perennial.

Kupchik, A. (2009). Things are tough all over: Race, ethnicity, class, and school discipline. Punishment &

Society, 11, 291–317.

Kupchik, A. (2010). Homeroom security: School discipline in an age of fear. New York: NYU Press.

Kupchik, A., & Monahan, T. (2006). The new American school: Preparation for post-industrial discipline.

British Journal of Sociology of Education, 27, 617–632.

Lareau, Annette. 2003. Unequal childhoods: Class, race, and family life. Berkeley, CA: University of California

Press.

Lawrence, R. (2006). School crime and juvenile justice. New York, NY: Oxford University Press.




Lawrence, R., & Mueller, D. (2003). School shootings and the man-bites-dog criterion of newsworthiness.

Youth Violence and Juvenile Justice, 1, 330–345.

Levine-Rasky, C. (2009). Dynamics of parent involvement at a multicultural school. British Journal of Sociology

of Education, 30, 331–344.

Lewis, A. E. (2006). Race in the schoolyard: Negotiating the color line in classrooms and communities. New

Brunswick, NJ: Rutgers University Press.

Lewis, R., Romi, S., Katz, Y. J., & Qui, X. (2008). Students’ reaction to classroom discipline in Australia,

Israel, and China. Teaching and Teacher Education, 24, 715–724.

Lohman, J., & Shephard, A. (2006). School security technologies (Office of Legislative Research Report (2006-

R-0668)). Hartford, CT: Connecticut General Assembly.

Lynch, M. P. (2009). Sunbelt justice: Arizona and the transformation of American punishment. Stanford, CA:

Stanford University Press.

Lyons, W., & Drew, J. (2006). Punishing schools: Fear and citizenship in American public education. Ann

Arbor, MI: University of Michigan Press.

Miller, L. (2008). The perils of federalism: Race, poverty, and the politics of crime control. New York, NY:

Oxford University Press.

Monahan, T., & Torres, R. (eds.). (2010). Schools under surveillance: Cultures of control in public education.

New Brunswick, NJ: Rutgers University Press.

Morris, E. W. (2005). ‘Tuck in that shirt!’ Race, class, gender, and discipline in an urban school. Sociological

Perspectives, 48, 25–48.

Mukherjee, E. (2007). Criminalizing the classroom: The over-policing of New York City schools. New York,

NY: New York Civil Liberties Union

Noguera, P. (2003). City schools and the American dream: Reclaiming the promise of public education.

New York, NY: Teachers College Press.

Nolan, K. (2011). Police in the hallway: Discipline in an urban high school. Minneapolis, MN: University of

Minnesota Press.

Payne, A. A., & Welch, K., (2010). Modeling the effects of racial threat on punitive and restorative school

discipline practices. Criminology, 48, 1019–1062.

Peguero, A. A. (2011). Immigration, schools, and violence: Assimilation and student misbehavior. Sociological

Spectrum, 31, 695–717.

Pickett, J., & Chiricos, T. (2012). Controlling other people’s children: Racialized views of delinquency and

whites’ punitive attitudes toward offenders. Criminology, 50, 673–710.

Renzulli, L. A., Parrott, H. M., & Beattie, I. R. (2011). Racial mismatch and school type: Teacher satisfaction

and retention in charter and traditional public schools. Sociology of Education, 84, 23–48.

Rist, R. C. (1973). The urban school: A factory for failure. Cambridge, MA: MIT Press.

Robers, S., Zhang, J., Truman, J., & Snyder, T. D. (2012). Indicators of school crime and safety: 2011.

Washington, DC: U.S. Department of Education and U.S. Department of Justice, Office of Justice Programs

(NCES 2012-002; NCJ 236021).

Simmons, L. (2007). The public school, the prison and the bottom line: Expenditure and expediency. Paper pre-

sented at American Sociological Association annual meeting, New York, NY.

Simon, J. (2007). Governing through crime: How the war on crime transformed American democracy and cre-

ated a culture of fear. New York, NY: Oxford University Press.

Skiba, R. J., Michael, R. S., Nardo, A. C., & Peterson, R. (2000). The color of ciscipline: Sources of racial

and gender disproportionality in school punishment. Indiana Education Policy Center, Research Report

SRS1.

Skiba, R., Reynolds, C. R., Graham, S., Sheras, P., Conoley, J. C., & Garcia-Vazquez, E. (2006). Are zero

tolerance policies effective in the schools? An evidentiary review and recommendations (Report by the

American Psychological Association Zero Tolerance Task Force). Washington, DC: American Psychological

Association.




Sweeten, G. (2006). Who will graduate? Disruption of high school graduation by arrest and court involvement.

Justice Quarterly, 23, 462–480.

United States Congress. (2000). H.R. 4108, Report No. 106-718. 106th Congress, 2d Session.

United States Department of Justice. (2011). Fiscal year 2011 secure our schools program. Retrieved April 19

2011, from http://www.cops.usdoj.gov/default.asp? Item¼2368

Unnever, J. D., & Cullen, F. T. (2010). The social sources of Americans’ punitiveness: A test of three competing

models. Criminology, 48, 99–130.

Wacquant, L. (2001). Deadly symbiosis: When ghetto and prison meet and mesh. In D. Garland (Ed.), Mass

imprisonment: Social causes and consequences (pp. 82–120). Thousand Oaks, CA: Sage.

Wald, J., & Losen, D. J. (eds.). (2003). Deconstructing the school-to-prison pipeline: New directions for youth

development. San Francisco, CA: Jossey-Bass

Webber, J. A. (2003). Failure to hold: The politics of school violence. New York, NY: Rowman and Littlefield.

Weis, L. (1992). Working class without work: High school students in a de-industrializing economy. New York,

NY: Routledge.

Welch, K., & Payne, A. A. (2010). Racial threat and punitive school discipline. Social Problems, 57, 25–48.

Welch, K., & Payne, A. A. (2012). Exclusionary school punishment: The effect of racial threat on expulsion and

suspension. Youth Violence and Juvenile Justice, 10, 155–171.

Welsh, W. N. (2003). Individual and institutional predictors of school disorder. Youth Violence and Juvenile

Justice, 1, 346–368.

Western, B. (2006). Punishment and inequality in America. New York, NY: Russell Sage Foundation.

Young, J. (1999). The exclusive society: Social exclusion, crime and difference in late modernity. Thousand

Oaks, CA: Sage.

Author Biographies

Aaron Kupchik is professor in the Department of Sociology and Criminal Justice at the University of

Delaware. He is author of Homeroom Security: School discipline in an age of fear (NYU Press 2010)

and Judging Juveniles: Prosecuting adolescents in adult and juvenile courts (NYU Press 2006).

Geoff Ward is associate professor in the Department of Criminology, Law & Society at the Univer-

sity of California, Irvine. He is author of The Black Child-Savers: Racial Democracy & Juvenile

Justice (University of Chicago Press 2012).



http://www.cops.usdoj.gov/default.asp? Item=2368

http://www.cops.usdoj.gov/default.asp? Item=2368


Documents

Youth Violence and Juvenile Justice Race, Poverty, and The